Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2604.08138 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866913022022254592 |
|---|---|
| author | Gogawale, Sharva Grudka, Gal Vasyutinsky-Shapira, Daria Ventura, Omer Kurar-Barakat, Berat Dershowitz, Nachum |
| author_facet | Gogawale, Sharva Grudka, Gal Vasyutinsky-Shapira, Daria Ventura, Omer Kurar-Barakat, Berat Dershowitz, Nachum |
| contents | A join is a set of manuscript fragments identified as originally emanating from the same manuscript. We study manuscript join retrieval: Given a query image of a fragment, retrieve other fragments originating from the same physical manuscript. We propose Bag of Bags (BoB), an image-level representation that replaces the global-level visual codebook of classical Bag of Words (BoW) with a fragment-specific vocabulary of local visual words. Our pipeline trains a sparse convolutional autoencoder on binarized fragment patches, encodes connected components from each page, clusters the resulting embeddings with per-image k-means, and compares images using set-to-set distances between their local vocabularies. Evaluated on fragments from the Cairo Genizah, the best BoB variant (viz. Chamfer) achieves Hit@1 of 0.78 and MRR of 0.84, compared to 0.74 and 0.80, respectively, for the strongest BoW baseline (BoW-RawPatches-$χ^2$), a 6.1% relative improvement in top-1 accuracy. We furthermore study a mass-weighted BoB-OT variant that incorporates cluster population into prototype matching and present a formal approximation guarantee bounding its deviation from full component-level optimal transport. A two-stage pipeline using a BoW shortlist followed by BoB-OT reranking provides a practical compromise between retrieval strength and computational cost, supporting applicability to larger manuscript collections. The code and dataset are available at https://github.com/TAU-CH/midrash_bob. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_08138 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Bag of Bags: Adaptive Visual Vocabularies for Genizah Join Image Retrieval Gogawale, Sharva Grudka, Gal Vasyutinsky-Shapira, Daria Ventura, Omer Kurar-Barakat, Berat Dershowitz, Nachum Computer Vision and Pattern Recognition A join is a set of manuscript fragments identified as originally emanating from the same manuscript. We study manuscript join retrieval: Given a query image of a fragment, retrieve other fragments originating from the same physical manuscript. We propose Bag of Bags (BoB), an image-level representation that replaces the global-level visual codebook of classical Bag of Words (BoW) with a fragment-specific vocabulary of local visual words. Our pipeline trains a sparse convolutional autoencoder on binarized fragment patches, encodes connected components from each page, clusters the resulting embeddings with per-image k-means, and compares images using set-to-set distances between their local vocabularies. Evaluated on fragments from the Cairo Genizah, the best BoB variant (viz. Chamfer) achieves Hit@1 of 0.78 and MRR of 0.84, compared to 0.74 and 0.80, respectively, for the strongest BoW baseline (BoW-RawPatches-$χ^2$), a 6.1% relative improvement in top-1 accuracy. We furthermore study a mass-weighted BoB-OT variant that incorporates cluster population into prototype matching and present a formal approximation guarantee bounding its deviation from full component-level optimal transport. A two-stage pipeline using a BoW shortlist followed by BoB-OT reranking provides a practical compromise between retrieval strength and computational cost, supporting applicability to larger manuscript collections. The code and dataset are available at https://github.com/TAU-CH/midrash_bob. |
| title | Bag of Bags: Adaptive Visual Vocabularies for Genizah Join Image Retrieval |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2604.08138 |